JP2001005978A - Device and method for extracting object, and program providing medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make easily extractable an object by discriminating that mutually adjacent measuring points are in exitence in a continuous surface area based on judgment that the distance between two points is smaller than a threshold value. SOLUTION: A continuous surface extracting part 401 scans the surface of a distance picture data displaying screen and obtains distance information of each element to execute judging and labeling of a continuous surface. Each pixel position is shown as coordinate points. The part 401 compares a distance between adjacent measuring points with a predetermined threshold value and when the distance is smaller than the threshold value, judges that the measuring point and its left adjacent measuring point are a continuous surface. An objective area selecting part 402 transforms labeling data to a distance picture to display it and executes the selecting processing of the object area. By presenting the picture obtained by transforming the labeling data to the distance picture, an operator easily selects the area of the same labeling to extract the object.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to an object extracting apparatus, an object extracting method, and a program providing medium recording the object extracting method. In particular, an object extraction device, an object extraction method, and a program recording an object extraction method that enable an object existing in a certain distance range in a three-dimensional space to be separated and extracted from other objects and the background and to perform extraction. Regarding the medium.

[0002]

2. Description of the Related Art Conventional methods for obtaining a three-dimensional shape of an object are roughly classified into an active method (active method) and a passive method (passive method). As an active method,
A method of calculating the distance to each measurement object by projecting light and measuring the time until the light is reflected back, or applying a slit-shaped pattern light to the measurement object and projecting it on the measurement object There is a method such as a light-section method for measuring a three-dimensional shape by examining the shape of the patterned light.

[0003] A typical passive method is as follows.
This is a stereo image method that uses the principle of triangulation, and is a method of measuring a distance by using two or more cameras to find a corresponding point between the images to obtain parallax.
For the stereo method, for example, “Minolta non-contact three-dimensional shape input device VIVID 700”, measurement and control, V
OL. 36, no. 12; PAGE. 910-911;
1997 or Kanade, Nakahara, Okutomi "Three-dimensional measurement by multi-baseline stereo method", Image Lab, pp
53-57.

In order to use the three-dimensional shape measured by the three-dimensional sensor as described above for applications such as modeling, only the area of the target object is cut out from the background and other objects from the measured three-dimensional information, ie, It is necessary to perform object extraction. Conventionally, as this object extraction method, by treating the distance information as a two-dimensional distance image, a region of the target object in the distance image is manually cut out using a commercially available image editing tool, for example, cut out using a mouse or the like. Or 3D-C
This is a method of manually cutting out unnecessary portions using G software, which requires a lot of time and labor.

[0005] Here, the distance image is a black-and-white grayscale image in which depth information is represented by a luminance value. As a means for assisting this, a clipping method using contour extraction by a snake model (Michael Kass, Andre)
w Witkn and Demeter Terzo
poulos "Snakes: Active Con
tour Models "[Internationa
l Journalnal Computer visi
on, pp. 321-331, 1988]), and using this function in a distance image can slightly reduce the amount of work. However, in the present method as well, the operation of specifying the vicinity of the contour of the target object by the operator is indispensable. That is, as shown in FIG. 1, when a specific clipping target is selected from a plurality of objects shown in the distance image 101, an area specifying frame indicating a surrounding area near the contour of the extraction object to specify the extraction object 102 An operation for generating 103 is required. The work of generating the area designation frame has a disadvantage that the work is complicated and time-consuming when the target object is complicated or when there are a plurality of target objects.

As an object extraction method other than the above-mentioned method,
There is a method of dividing a section in the distance direction and extracting only the objects in the section.However, this method also requires an operator to adjust the distance section, which increases the number and complexity of the objects. There is no doubt that the adjustment work will increase.

In the method of dividing the distance section, there is a case where it is not possible to cut out only one object depending on the arrangement of the objects. This example will be described with reference to FIGS.

As shown in FIG. 2, in front of a wall 203, there are two measurement objects, measurement objects 1,201 and measurement objects 2,2.
It is assumed that a three-dimensional sensor is arranged in the space where 02 is arranged as shown in FIG. 2 to obtain depth information. FIG. 3A shows a distance image obtained by the three-dimensional sensor.

When it is desired to cut out the measurement objects 1,201 and 2,202 from the distance image shown in FIG. 3A, the distance section shown in FIG. Can be cut out. However,
When it is desired to cut out only the measurement target 1 from here, the distance image when the distance section in FIG. 2 is set to the distance section 1 is as shown in FIG. Can not. In addition, the distance section in FIG.
3C is as shown in FIG. 3C, and not only the measurement targets 1, 201 but also the measurement targets 2, 20
As a result, a part of 2 is also cut out together. The distance image when the distance section of FIG. 2 is set to the distance section 3 is (d) of FIG.
, Measurement objects 1,201, measurement objects 2,20
2 are cut out together.

[0010]

As described above, in the conventional object extraction method, for example, when a target object is extracted by using an image editing tool in a distance image based on measured three-dimensional information, a target object area is extracted. Is a manual operation using a mouse or the like, resulting in a result requiring a lot of time and labor. Also, in a method of dividing a section in the distance direction and extracting only an object in the section,
It is necessary for the operator to adjust the distance section, and the setting work of the optimum distance section becomes complicated according to the number and complexity of the objects, resulting in an increase in the adjustment work.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned drawbacks of the conventional object extraction method, and has an object extraction apparatus for efficiently extracting an object from three-dimensional information obtained from a three-dimensional sensor. It is an object to provide an object extraction method and a program providing medium.

According to the present invention, three-dimensional information is obtained from a three-dimensional sensor by identifying measurement points existing within a certain distance range, for example, a distance equal to or less than a threshold in a three-dimensional space as continuous surfaces. For example, object extraction can be easily performed by dividing the range image that displays the original information into regions, selecting one or more surfaces from the identified continuous surface regions, and deleting the other surface regions It is an object to provide a simple object extraction device, an object extraction method, and a program providing medium.

[0013]

Means for Solving the Problems The present invention has been made in consideration of the above problems, and a first aspect of the present invention is based on three-dimensional information. In the object extracting device that performs the extraction, the distance information between two points between two adjacent measurement points among the measurement points included in the three-dimensional information is calculated, and the calculated distance information between the two points and a predetermined threshold value are calculated. A continuous surface extracting means for performing the comparison process of the above, and identifying the adjacent measurement points as a continuous surface region having continuity based on the determination that the distance information between the two points is smaller than the threshold value. An object extraction device characterized by having:

Further, in one embodiment of the object extracting apparatus according to the present invention, the distance information between the two points is calculated by a three-dimensional coordinate calculated from three-dimensional coordinates of the two adjacent measurement points obtained based on the three-dimensional information. It is a characteristic distance.

Further, in one embodiment of the object extracting apparatus according to the present invention, the distance information between the two points is calculated from three-dimensional coordinates of the two adjacent measurement points obtained based on the three-dimensional information. It is a depth direction distance that is an axial distance.

Further, in one embodiment of the object extracting apparatus of the present invention, the continuous surface extracting means has a configuration for executing a labeling process for giving a label for identifying a different continuous surface region to the measurement point. In addition, a feature is provided in which a labeling process for giving the same label to each of two measurement points identified as a continuous surface region having continuity is performed.

Further, in one embodiment of the object extracting apparatus according to the present invention, the image is a distance image obtained based on three-dimensional information, and the continuous plane extracting means is provided for measuring points included in the distance image. Determining the continuity between the measurement point and the upper adjacent measurement point based on a comparison process between the two points, the distance information DU between the measurement point and the upper adjacent measurement point adjacent in the upward direction, and the threshold value; The continuity between the measurement point and the left adjacent measurement point is determined based on a comparison process between the point-to-point distance information DL between the measurement point and the left adjacent measurement point adjacent to the left direction and the threshold, and Point distance information DL, D
If only one of U is smaller than the threshold,
If the same label as the adjacent measurement point smaller than the threshold is given as the label of the measurement point, and if any of the distance information between two points DL and DU is not smaller than the threshold,
A new label is given as a label of the measurement point, and when the distance information between two points DL and DU are both smaller than the threshold value, one of the identification labels in the upward direction and the left direction is replaced with the label. A process of changing the label of the measurement point to which the same label as the non-selected label in one of the upward and left directions different from the selected label is added to the selected label and the label of the measurement point is changed to the selected label. Is performed.

Further, in one embodiment of the object extracting apparatus according to the present invention, the image is a distance image obtained based on three-dimensional information, and the continuous plane extracting means is provided for measuring points included in the distance image. The distance information DU between two points with the upper adjacent measurement point adjacent to the measurement point in the upper direction, the distance information DL between the two points on the upper adjacent measurement point adjacent to the measurement point in the left direction, the downward direction of the measurement point Based on comparison processing between each of the two-point distance information DD with the upper adjacent measurement point adjacent to the measurement point and the two-point distance information DR with the upper adjacent measurement point adjacent to the measurement point rightward. Determine the continuity between the measurement point and each of the adjacent measurement points in the four directions,
With the adjacent measurement point determined to have continuity as the next measurement point, a continuous plane extraction is performed by a four-connection method in which the process of determining continuity with each of the adjacent measurement points in four directions, up, down, left, and right, is sequentially repeated. It is characterized by having.

Further, in one embodiment of the object extracting apparatus according to the present invention, the continuous plane extracting means comprises: for measuring points included in a distance image obtained based on three-dimensional information, eight adjacent measuring points around the measuring point; Continuity surface extraction by the 8 concatenation method that repeats the process of successively determining the continuity with the neighboring measurement points determined to be continuous as the next measurement point and successively determining the continuity with eight neighboring measurement points Is performed.

Further, in one embodiment of the object extracting apparatus according to the present invention, the continuous plane extracting means sets an arbitrary point of an image representing the object to be extracted as a starting point of the measuring point, and sequentially connects the continuous point with an adjacent measuring point. And performs continuous surface extraction.

Further, in one embodiment of the object extracting apparatus of the present invention, each of the measurement points is associated with a pixel included in the image, and an adjacent measurement point corresponds to an adjacent pixel. It is characterized by.

According to a second aspect of the present invention, there is provided an object extraction method for extracting an object constituting a continuous surface area from an image based on three-dimensional information. A point-to-point distance information calculating step of calculating point-to-point distance information between two adjacent measurement points; a comparing step of executing a comparison process between the calculated point-to-point distance information and a predetermined threshold; A step of identifying the adjacent measurement points as a continuous plane area having continuity based on the determination that the distance information between the two points is smaller than the threshold value. In the object extraction method.

Further, in one embodiment of the object extraction method of the present invention, the two-point distance information used in the two-point distance information calculating step is the two adjacent measurement points obtained based on the three-dimensional information. Is a three-dimensional distance calculated from the three-dimensional coordinates.

Further, in one embodiment of the object extraction method of the present invention, the distance information between two points used in the step of calculating distance information between two points is the distance between the two adjacent measurement points obtained based on the three-dimensional information. , Which is a distance in the depth direction, which is a distance in the Z-axis direction calculated from the three-dimensional coordinates.

Further, in one embodiment of the object extraction method of the present invention, the object extraction method further comprises a labeling processing step of giving a label for identifying a different continuous surface area to the measurement point, In the labeling processing step, the same label is assigned to each of the two measurement points identified as the continuous plane area having continuity in the continuous plane area identification step.

Further, in one embodiment of the object extracting method of the present invention, the step of calculating the distance information between two points includes the step of calculating the distance between the measurement points included in the distance image obtained based on the three-dimensional information. Calculating the distance information DU between two points with the upper adjacent measurement point to be measured, and calculating the distance information DL between two points between the measurement point and the left adjacent measurement point adjacent in the left direction, The comparing step includes a step of comparing the distance information between two points DU and the threshold value, and a step of comparing the distance information between two points DL and the threshold value. The region identification step includes the measurement point, the left adjacent measurement point, and the upper adjacent measurement point based on the comparison result in the comparison processing step. With determining the continuity with,
When only one of the two-point distance information DL and DU is smaller than the threshold, the same label as an adjacent measurement point smaller than the threshold is given as the label of the measurement point, and the two-point distance is calculated. If none of the information DL and DU is smaller than the threshold value, a new label is given as the label of the measurement point, and the two-point distance information D
When both L and DU are smaller than the threshold value, any one of the identification labels in the upward direction and the left direction is used as a selection label, and the label is assigned as a label of the measurement point, and is different from the selection label. A process of changing the label of the measurement point to which the same label as the non-selected label in the upward direction or the left direction is added to the selected label is executed.

Further, in one embodiment of the object extraction method of the present invention, the step of calculating the distance information between two points includes the step of calculating the distance between the measurement points included in the distance image obtained based on the three-dimensional information. Distance information DU between two points with an adjacent upper adjacent measurement point, distance information DL between two points with an upper adjacent measurement point adjacent to the left of the measurement point, and an upper adjacent measurement point adjacent below the measurement point And the two-point distance information DD with an upper adjacent measurement point adjacent to the measurement point in the right direction of the measurement point, and the comparison processing step is performed in the two-point distance information calculation step. Performing a comparison process between each of the calculated two-point distance information DU, DL, DD, and DR and the threshold value; Determining the continuity of the upper and lower left and right adjacent measuring point and the measurement point on the basis of the comparison result of physical step, the it is determined that there is continuity adjacent measuring points as the next measurement point, further vertical and horizontal 4
The method is characterized in that continuous surface extraction is performed by a four-connection method in which processing for determining continuity with each adjacent measurement point in the direction is sequentially repeated.

Further, in one embodiment of the object extraction method according to the present invention, the measurement points included in the distance image obtained based on the three-dimensional information are set to be around the measurement points.
An 8 concatenation method in which the continuity with two adjacent measurement points is determined, the adjacent measurement point determined to have continuity is set as the next measurement point, and the process of determining the continuity with eight adjacent measurement points is sequentially repeated. , The continuous plane extraction is performed by

Further, in one embodiment of the object extraction method of the present invention, the method further comprises a starting point setting step of setting an arbitrary point of the image showing the object to be extracted as a starting point of the measurement point, and calculating the distance information between two points. The step is characterized in that the points set in the starting point setting step are set as the first measurement points, and the continuity with the adjacent measurement points is sequentially determined to perform the continuous plane extraction.

In one embodiment of the object extraction method according to the present invention, each of the measurement points is associated with a pixel included in the image, and an adjacent measurement point corresponds to an adjacent pixel. It is characterized by.

Further, a third aspect of the present invention is a program providing medium tangibly providing a computer program for causing a computer system to execute a process for executing an object extraction process based on three-dimensional information. The computer program further comprises: a point-to-point distance information calculating step of calculating point-to-point distance information between two adjacent measurement points among the measurement points included in the three-dimensional information; and a calculated point-to-point distance. A comparison processing step of performing comparison processing between information and a predetermined threshold value, and a continuous surface in which the adjacent measurement points have continuity based on the determination that the distance information between the two points is smaller than the threshold value. And a continuous surface area identifying step of identifying the area as an area.

The program providing medium according to the third aspect of the present invention is, for example, a medium for providing a computer program in a computer-readable format to a general-purpose computer system capable of executing various program codes. . The form of the medium is not particularly limited, such as a storage medium such as a CD, an FD, and an MO, and a transmission medium such as a network.

Such a program providing medium defines a structural or functional cooperative relationship between the computer program and the providing medium for realizing the functions of a predetermined computer program on a computer system. Things. In other words, by installing the computer program into the computer system via the providing medium, a cooperative operation is exerted on the computer system, and the same operation and effect as those of the first aspect of the present invention are obtained. You can do it.

[0034] Still other objects, features and advantages of the present invention are:
It will become apparent from the following more detailed description based on the embodiments of the present invention and the accompanying drawings.

[0035]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention eliminates the problems of the conventional object extraction method, that is, the complexity of cutting out a target object region or adjusting a distance section in a distance image based on three-dimensional information and improving efficiency. An object extraction device, an object extraction method, and a program providing medium that enable good object extraction. The three-dimensional information obtained from a three-dimensional sensor is used to continuously connect measurement points existing in a certain distance range on a three-dimensional space. Are divided into areas as
In this configuration, an object is extracted by selecting one or a plurality of planes and deleting other plane areas. Less than,
An object extraction device, an object extraction method, and a program providing medium of the present invention will be described in detail with reference to the drawings.

[Embodiment 1] The first embodiment of the object extracting apparatus of the present invention
In the embodiment, based on three-dimensional shape data obtained from a three-dimensional sensor, a continuous surface area is identified, the same label is assigned to measurement points determined to be continuous, and different measurement points are determined to be different surfaces. In this configuration, labeling for giving a label is performed, and an operator selects one or a plurality of target objects from an area divided into a plurality of surface regions by the labeling, thereby extracting an object, that is, extracting the target object.

FIG. 4 shows a first example of the object extracting apparatus according to the present invention.
FIG. 2 shows a functional block diagram of the embodiment. FIG. 9 shows an example of a hardware configuration of the object extraction device of the present invention, which will be described later in detail. The object extracting apparatus according to the present embodiment includes a continuous plane extracting unit 401 and a target area selecting unit 402 as shown in FIG.

The continuous plane extracting unit 401 receives three-dimensional shape data or a distance image obtained from a three-dimensional sensor,
A continuous surface is identified from the three-dimensional shape data or the distance image, and labeling for assigning different labels to the plurality of identified continuous surface regions is executed.

As the three-dimensional shape data or the distance image received by the continuous plane extracting unit 401, the same three-dimensional information data as described in the section of the prior art can be used. The data obtained. For example, three-dimensional shape data and range image data obtained by a passive method (passive method) such as a stereo image method using the principle of triangulation are used.

The principle of the stereo image method will be briefly described. The stereo image method uses two or more cameras.
The position of a measurement target in a three-dimensional space is to be obtained by associating pixels in a plurality of images obtained by photographing the same target from more than one direction (different line-of-sight directions). For example, the same object is photographed from different directions by the reference camera and the reference camera, and the distance between the measurement objects in each image is measured based on the principle of triangulation.

FIG. 5 is a diagram for explaining the principle of the stereo image method. The reference camera and the reference camera photograph the same object from different directions. Consider obtaining the depth of a point “mb” in an image captured by a reference camera.

The object seen at the point "mb" of the image taken by the reference camera is indicated by "m1,""m2," and "m3" in the image taken by the reference camera taking the same object from different directions. Will be developed on a straight line. This straight line is called an epipolar line.

The position of the point "mb" in the reference camera appears on a straight line called "epipolar line" in the image taken by the reference camera. Point P to be imaged
(Points existing on a straight line including P1, P2, and P3) are the same observation point “mb” on the reference image regardless of the depth, that is, the distance from the reference camera, as long as they are on the line of sight of the reference camera. Appears in On the other hand, the point P on the image captured by the reference camera appears on the epipolar line at a position corresponding to the magnitude of the distance between the reference camera and the observation point P.

FIG. 5 illustrates the correspondence between the epipolar line and the observation point “mb” in the reference image. As shown in the figure, as the position of the observation point P changes to P1, P2, and P3, the observation point in the reference image is “m1”,
Shift to “m2” and “m3”.

By using the above geometrical optical properties to search for the observation point "mb" on the epipolar line, the distance of the point P can be identified. This is the basic principle of the “stereo image method”. In this way, three-dimensional information on all pixels on the screen is obtained.
The acquired three-dimensional information can be used as pixel attribute data corresponding to each pixel.

Although the above-described stereo image method uses a single reference camera and a single reference camera, a multi-baseline stereo (Multi) using a plurality of reference cameras is used.
An evaluation value may be obtained by an iBaseline Stereo) method, and three-dimensional information for each pixel may be obtained based on the evaluation value. The multi-baseline stereo image method uses images taken by one reference camera and a plurality of reference cameras, obtains an evaluation value representing a correlation between each of the plurality of reference camera images and the reference camera image, and calculates each evaluation value. The sum is added and the sum is used as a final evaluation value. For details of the multi-baseline stereo image method, see, for example, “Stereo matching using a plurality of baseline lengths”, IEICE Transactions D
-11Vol. J75-D-II No. 8 pp. 1
317-1327, August 1992.

Note that the above-described stereo image method is only one method for obtaining three-dimensional information, and three-dimensional shape data and distance information used in the object extracting apparatus of the present invention are obtained by this stereo image method. Not limited to data, for example, data obtained by measuring a light reflection time from an object, or a light cutting method by examining a shape of a slit-shaped pattern light projected on a measurement target, and may include distance information of each object. Any data that can be used.

The continuous surface extraction unit 401 in the object extraction device of the present invention performs a continuous surface extraction process based on the received three-dimensional shape data. This continuous plane extraction processing will be described with reference to FIG.

The three-dimensional shape data received by the continuous plane extracting unit 401 can be expressed as a distance image viewed from the three-dimensional sensor used when generating the three-dimensional shape data. The distance image is image data in which three-dimensional data of a display element in each pixel on the display image is displayed in association with each pixel.

As shown in FIG. 6A, the continuous plane extracting unit 401 scans the distance image data display screen from the upper left to the right as shown in FIG. 6A, and when the distance image reaches the right end of one line, the next line is extracted. Is performed, the distance information of each pixel is obtained, the continuous plane is determined, and the labeling is performed. Each pixel position is a coordinate point P where i is the scan line direction and j is the direction orthogonal to the scan line.
Shown as (i, j).

FIG. 6B shows a part of the data obtained by developing the distance image into a three-dimensional space. In FIG. 6B, a distance DL between a certain measurement point P (i, j) in the distance image and an adjacent measurement point P (i-1, j) on the left side thereof.
Ask for. The three-dimensional coordinates of the measurement point P (ij) are represented by (x
(I, j), y (i, j), z (i, j)), and the three-dimensional coordinates of the measurement point P (i-1, j) are (x (i−
(1, j), y (i-1, j), z (i-1, j)), the measurement point P (i, j) and the adjacent measurement point P (i-1, j) on the left side thereof Is determined by the following equation (1).

[0052]

(Equation 1)

The continuous plane extracting unit 401 compares the distance DL between the measurement point obtained by the above equation and the measurement point adjacent to the left with a predetermined threshold value TH, and calculates the distance D
When L is smaller than the threshold value TH, the measurement point P (i,
j) and an adjacent measurement point P (i−1,
j) is determined to be a continuous surface.

The threshold value TH used in the continuous plane extracting unit 401 is, for example, the length of a diagonal line of a cube whose one side is the largest value in the XYZ directions of the resolution of the three-dimensional sensor, or the size of the depth resolution. It can be used, and the operator can set an arbitrary threshold. These thresholds are set in advance, and the continuous surface extraction unit 4
01 is stored in the accessible storage means.

Next, the continuous plane extraction unit 401 determines the measurement point P (i, j) and the adjacent measurement point P
The distance DU from (i, j-1) is obtained. Measurement point P
Let the three-dimensional coordinates of (ij) be (x (i, j), y (i,
j), z (i, j)) and the measurement point P (i, j−
The three-dimensional coordinates of (1) are expressed as (x (i, j-1), y (i, j-
1), z (i, j-1)), the measurement point P
(I, j) and the adjacent upper measuring point P
The distance DU to (i, j-1) is obtained by the following equation (2).

[0056]

(Equation 2)

The continuous surface extraction unit 401 compares the distance DU obtained by the above equation with the threshold TH, and when the distance DU is smaller than the threshold TH, the measurement point P (i, j)
And an adjacent upper measuring point P (i, j-1)
Is determined to be a continuous surface.

The continuous plane extraction unit 401 calculates
The labeling process is performed using the two distances DL and DU. The labeling process is determined as follows based on the continuity and discontinuity of adjacent measurement points determined in the comparison between DL and DU and the threshold value TH.

(1) In the case of left continuation and top continuation, * DL <TH and DU <TH, the label number L (i−j) of the left point P (i−1, j)
(1, j) is compared with the label number L (i, j-1) of the upper point P (i, j-1), and the smaller label number is used as the label number of the measurement point P (i, j). .

(2) In the case of left continuous and upper discontinuous: * DL <TH and DU ≧ TH, label number L (i−j) of left point P (i−1, j)
(1, j) is the label number of the measurement point P (i, j).

(3) In the case of left discontinuity and upper continuous, * DL ≧ TH and DU <TH, the label number L (i, j) of the upper point P (i, j-1)
j-1) is the label number of the measurement point P (i, j).

(4) In the case of left discontinuity and upper discontinuity: * DL ≧ TH and DU ≧ TH, label number L (i, j) of current measurement point P (i, j)
Substitute the new number "Label" for "Label"
Is incremented by one.

The labeling of each measurement point is executed based on the labeling principles (1) to (4).

FIG. 7 shows a flow of the labeling process in the continuous plane extracting unit 401. Hereinafter, each step of FIG. 7 will be described.

Step 701 shows an initialization process. The measurement point is a point P (i,
j) the scanning direction from the upper left point P (0, 0),
That is, j is incremented by 1 at the point where it proceeds in the i increasing direction and reaches the right end, that is, the point of i = MAX, and the next line is measured. In the initialization step 701, i = 0,
j = 0, label = 0, and L (0, 0) = 0 are set.

In step 702, it is determined whether or not the i coordinate of the measurement point is 0. In steps 703 and 704, j is determined.
It is determined whether the coordinates are 0 or not. The point of i = 0 means the point of the top line, and the distance DU to the above-mentioned upper adjacent point cannot be obtained. Also, the point at j = 0 means the leftmost point, and the distance DL from the above-described left adjacent point cannot be obtained. At other points, DL and DU can be obtained. As described above, the distance information that can be calculated differs depending on the position of the measurement point.

Therefore, steps 705 to 707 are executed according to these three modes. Steps 705 to 707
Then, the distance DL and the distance DU value are calculated, or the distance D
L, at points where distance DU value cannot be measured, DL,
The value of DU is set to a value equal to the threshold value TH.

Steps 708 to 710 are steps for performing classification by comparing the distances DL and DU with the threshold value TH. If both steps 708 and 709 are Yes, the above (1) left continuation and top continuation,
DL <TH and DU <TH, and in the next step 711, the left point P (i−1,
j) is compared with the label number L (i, j-1) of the upper point P (i, j-1), and in the next step 712 or 713, Use the smaller label number as the measurement point P (i, j)
Label number.

Step 708 is Yes, Step 709
Is No, the above-mentioned (2) left-continuous, upper-discontinuous corresponds to the case of DL <TH and DU ≧ TH, and in step 714, the left point P (i−1, j)
The label number L (i-1, j) of the measurement point P (i, j)
j) label number.

If step 708 is No and step 710 is Yes, the above (3) left discontinuity and upper continuation correspond to the case of DL ≧ TH and DU <TH. Point P (i, j-1)
Label number L (i, j-1) of measurement point P (i, j-1)
j) label number.

If the step 708 is No and the step 710 is No, the above-mentioned (4) case of left discontinuity and upper discontinuity corresponds to the case of DL ≧ TH and DU ≧ TH. The new number “Label” is substituted for the label number L (i, j) of the measurement point P (i, j), and the value of “Label” is increased by one.

After these processes, it is determined in step 717 whether i has reached the maximum value, and in step 718 whether j has reached the maximum value.
At 9,720, the i value and the j value are increased by 1, and labeling of the next measurement point is performed.

By executing the processing flow as described above up to the final measurement point: P (MAX, MAX), labels at all measurement points are determined.

However, since the label numbers obtained in this processing flow are not arranged in order from the largest one or the smallest one, it is necessary to assign a label number according to the distance. It is necessary to check the correspondence between the set label number and the area and convert the label number.

In the target area selection section 402 in FIG. 4, a process of selecting a target area by an operator is performed. The target area selection unit 402 converts the labeling data obtained as a result of the completion of the labeling processing into a distance image and displays the distance image on the display unit. FIG. 8 shows an example of an image obtained by converting labeling data into a distance image. FIG. 8 shows three-dimensional data obtained by the object arrangement and the three-dimensional sensor arrangement shown in FIG.
Is an image obtained by converting labeling data obtained when the labeling process flow of FIG.

In the area shown in FIG. 8, label 0 corresponds to the background. The label 1 corresponds to the wall 203 in FIG. 2, the label 2 corresponds to the measurement objects 1, 201, and the label 3 corresponds to the measurement objects 2, 202. By presenting an image obtained by converting the labeling data into a distance image as shown in FIG. 8 to the operator, the operator can easily select the same labeled region, and as a result, the object is extracted. For example, when selecting one or a plurality of regions to be cut out,
By performing a process of selecting one or a plurality of different label regions and deleting the other label regions, an object to be selected can be easily extracted. Note that the operator can designate a specific label area, and then select the extracted area to be displayed three-dimensionally.

FIG. 9 shows an example of a hardware configuration of the object extracting apparatus according to the present embodiment. The example of the hardware configuration shown in FIG. 9 is a computer device 90 such as a personal computer.
1, a display unit 908, a mouse 909 as an input device, and a keyboard 910. The computer device 901 includes a CPU 902, a ROM 903,
RAM 904, display interface 905, input device interface 906, H as a storage device
DD907.

The CPU 902 is a main controller that controls various processing operations as the object extracting apparatus, and starts a predetermined application program for executing pixel selection in an image, comparison processing, and other various processing of the image. Unit. The CPU 902 performs control in the processing of the continuous plane extraction unit 401 and the target area selection unit 402 shown in FIG. 4 according to the respective processing programs.

The ROM 903 stores permanent program data such as an IPL, a self-diagnosis program (POST),
Code group for executing hardware input / output operation (B
IOS) and the like are stored.

The RAM 904 is a writable memory used by the CPU 902 to load an execution program code and temporarily store work data. It is used to temporarily store necessary data, programs, and the like in continuous surface extraction and target area selection processing.

The display interface 905 includes:
The display of the display unit 908 is controlled. Although not shown in FIG. 8, for example, the image data processed under the control of the CPU 902 is temporarily stored in the VRAM and supplied to the display unit 908.

The input device interface 906 is an interface device for receiving data, command input, and the like input from the user via the mouse 909 and the keyboard 910.

Hard Disk Drive (HDD) 9
Reference numeral 07 denotes a relatively large-capacity rewritable and non-volatile storage device, which is used for storing data files and installing program files.
Data files include captured image files,
3D shape data corresponding to image files, distance images,
Labeling data. Examples of the program file include a continuous plane extraction process executed by the CPU 902,
Labeling processing, other image synthesis, range image display application, CG application, and the like. In addition,
As the external storage device, various types of storage devices such as an optical disk and a magneto-optical disk can be used in addition to the HDD.

The display unit 908 is a device for presenting a work screen, such as a processed image and a menu for inputting a command, to a user, and is a CRT (Cathode Ray T).
ube: cold cathode ray tube) display and LCD (Liqu
id Crystal Display (Liquid Crystal Display). The display unit 908 performs display of a distance image, display of a three-dimensional image, display of an image obtained by converting labeling data into a distance image, and the like.

The input from the user is made by a mouse 909 for inputting a command in a coordinate designation format and a keyboard 910 for inputting a command on a character basis.

In order to actually configure the object extracting apparatus of the present invention, many hardware components other than those shown in FIG. 9 are required. However, these are well known to those skilled in the art and do not constitute the gist of the present invention, and thus are omitted in this specification. It should be noted that connections between hardware blocks in the drawings are also shown in an abstract form in order to avoid complicating the drawings.

[Second Embodiment] In the first embodiment described above, a configuration in which a three-dimensional distance between adjacent measurement points is determined when a continuous plane is determined has been described. However, a method for executing a continuous plane extraction by a simpler method has been described. A configuration in which a difference in only the depth direction between adjacent measurement points is determined and a continuous surface is determined based on the difference will be described as a second embodiment of the object extraction device of the present invention.

The functional block diagram and the hardware configuration of the object extracting apparatus according to the second embodiment are the same as those of the first embodiment, and the second embodiment will be described with reference to these figures. The continuous surface extraction unit 401 of the object extraction device according to the second embodiment calculates a difference only in the depth direction between adjacent measurement points.

The continuous plane extracting unit 401 is the same as in the first embodiment.
The three-dimensional shape data that can be expressed as a distance image is received, and the distance image is scanned from the upper left to the right on the distance image data as shown in FIG. A certain measurement point P (i, j) in the distance image
And the adjacent measurement point P (i-1, j) on the left side
Is the distance in the depth direction, that is, FIG.
It is obtained as the distance in the Z-axis direction in (b). Distance DL
Is determined by the following equation (3).

[0090]

(Equation 3)

The continuous plane extracting unit 401 compares the distance DL obtained by the above equation with a predetermined threshold TH. If the distance DL is smaller than the threshold TH, the measurement point P (i, j) and The adjacent measurement point P on the left
It is determined that (i-1, j) is a continuous surface.

The above-mentioned threshold TH can be, for example, the depth resolution of the three-dimensional sensor, and an operator can set an arbitrary threshold. These threshold values are set in advance, and are stored and held in an accessible storage unit of the continuous plane extraction unit 401.

Next, the continuous plane extraction unit 401 determines the measurement point P (i, j) and the adjacent upper measurement point P (i, j).
The distance DU from (i, j-1) is determined as the distance in the depth direction, that is, the distance in the Z-axis direction in FIG. 6B.
The distance DU is obtained by the following equation (4).

[0094]

(Equation 4)

The continuous plane extracting section 401 compares the distance DU obtained by the above equation with the threshold TH, and when the distance DU is smaller than the threshold TH, the measurement point P (i, j)
And an adjacent upper measuring point P (i, j-1)
Is determined to be a continuous surface.

The continuous plane extracting unit 401 calculates the distance D
A labeling process is performed using L and DU. The labeling process is performed based on the continuity and discontinuity of adjacent measurement points determined in the comparison between DL and DU and the threshold value TH, and is in accordance with the process flow described in the first embodiment and illustrated in FIG. It is.

The object extracting apparatus according to the second embodiment can obtain approximately the same effect as that of the first embodiment without performing expansion into a three-dimensional space. Therefore, all processes such as region division on a continuous plane, region selection by an operator, and the like can be executed using the distance image, and the amount of calculation is reduced and the user interface is simplified.

[Embodiment 3] In the above-described Embodiments 1 and 2, all measurement points, that is, from P (0,0) to P (MAX, MAX) are labeled, and the results are used. Although the configuration that performs object extraction by cutting out a continuous area was explained, just by specifying a part of the object to be extracted with a pointing device such as a mouse,
A configuration for cutting out a continuous region of the object will be described as a third embodiment of the object extraction device of the present invention.

The functional block diagram and the hardware configuration of the object extracting apparatus according to the third embodiment are the same as those of the first and second embodiments, and the third embodiment will be described with reference to these figures. The continuous surface extraction unit 402 of the object extraction device according to the third embodiment receives position data indicating a part of an object to be extracted specified by an operator.
The operator can use a pointing device such as a mouse 909 shown in FIG. 9 to display a specific point P from an image displayed on the display unit, for example, three-dimensional image data or a distance image.
(I, J) is specified.

The continuous plane extracting section 402 extracts only the continuous area to which the point P (I, J) instructed belongs. FIG. 10 shows a part of data obtained by developing the three-dimensional measurement data in the XYZ space. The point P (i,
If j) is the point P (I, J) specified by the worker, the continuous plane extraction unit 402
With j) as a starting point, the continuity between the point P (i, j) and adjacent measurement points in four directions, up, down, left, and right, is checked, and continuity is checked while sequentially moving the measurement points, and only continuous points are extracted. This is a configuration in which the continuity check is further performed on the extracted continuous points with the points in the surrounding four directions, which is a labeling using a so-called 4-connection method.

FIGS. 11 and 12 show a labeling processing flow executed by the continuous plane extracting unit 402 in the object extracting apparatus according to the third embodiment.

Step 1101 in FIG. 11 is a step for initially setting the measurement point P (i, j) to the user's designated point P (I, J). This is a step of executing a 4-connection check routine for checking continuity with adjacent measurement points in the four directions. Details of the 4-connection check routine are as shown in FIG.

The four connection check routine shown in FIG. 12 will be described. Steps 1201 to 1206 correspond to a measurement point P (i, j) and a point P (i, j) adjacent thereto.
j-1) is checked, and steps 1207 to 1212 check the continuity between the measurement point P (i, j) and a point P (i-1, j) adjacent to the measurement point P (i, j). Steps 1213 to 1218
Is a step of examining the continuity between the measurement point P (i, j) and a point P (i, j + 1) adjacent thereunder.
(I, j) and a point P (i + 1,
j) is a step of investigating continuity with j).

Steps 1201 to 1206 will be described. In step 1201, it is determined whether the point P (i, j-1) adjacent to the measurement point P (i, j) is unlabeled. If a label has already been assigned, it is determined that the number is N, and the next step, step 120, is performed.
Go to 7. If not, the process proceeds to step 1202, where it is determined whether or not the upper point P (i, j-1) is within the measurement range. For example, when the measurement point P (i, j) is the upper end line of the acquired image, there is no data at a point further above and it is determined that the point is not within the measurement range (ie, N), and the next step, step 1207
Proceed to.

If the upper point P (i, j-1) is unlabeled and is within the measurement range, the process proceeds to step 1203, where the point P (i, j) adjacent to the measurement point P (i, j) is located above. , J−1) is determined.

This continuity determination is made based on whether or not the distance DU obtained by the following equation (5) is smaller than a threshold value TH. The distance DU is the measurement point P (i
j) is defined as (x (i, j), y (i, j), z
(I, j)), and the above adjacent point P (i, j-1)
Are represented by (x (i, j-1), y (i, j-
1), z (i, j-1)), is obtained by the following equation (5).

[0107]

(Equation 5)

The continuous plane extraction unit 401 compares the distance DU obtained by the above equation with a predetermined threshold TH, and when the distance DU is smaller than the threshold TH, the measurement point P (i, j) and Adjacent measurement point P above
It is determined that (i, j-1) is a continuous surface (Yes in step 1203), and the process proceeds to step 1204, where the label L (i, j) of the upper adjacent measurement point P (i, j-1) is determined.
j-1) is changed to the label L (i, j) of the measurement point P (i, j).
Same as j).

As the threshold value TH, for example, the length of a diagonal of a cube having one side as the maximum value in the XYZ directions of the resolution of the three-dimensional sensor can be used, or the depth resolution can be used. It is also possible for an operator to set an arbitrary threshold. These thresholds are set in advance,
It is stored and held in an accessible storage unit of the continuous surface extraction unit 401.

Further, at step 1203, the measurement point is moved upward, that is, to P (i, j-1), and at step 1206, a similar 4-connection check routine is executed. Here, comparison of P (i, j-1) with the upper, lower, left, and right measurement points is performed.

In steps 1201 and 1202, if a label has already been assigned to an upper adjacent point, or if it is determined that the upper adjacent point is not within the measurement range, the process proceeds to step 1207, where the measurement point P (i , J) and the left point P (i-1, j) are determined. Hereinafter, similarly, Step 1213
In steps 1219 to 1224, the continuity between the measurement point P (i, j) and the adjacent point P (i, j + 1) is determined.
(I, j) and a point P (i + 1,
j) is determined. These determination processes are the same as steps 1201 to 1206, and a description thereof will be omitted.

In step 1209 which is a step for determining the continuity between the measurement point P (i, j) and the left point P (i-1, j), the distance DL of the following equation (6) is used.
Is compared with the threshold value TH. When the distance DL is smaller than the threshold value TH, it is determined that the distance is continuous. In comparison with the lower measurement point, the distance DD and the threshold T in Expression (7) are used.
In the comparison with H and the comparison with the right measurement point, the comparison between the distance DR of Expression (8) and the threshold value TH is performed.

[0113]

(Equation 6)

The flow shown in FIG.
01, 1202, 1207, 1208, 1213, 12
When all the determinations of 14, 12, 19, and 1220 are N, the process ends. By executing this 4-connection check routine, the user first specifies and sets the starting point P
Only a continuous region including (I, J) is extracted.

In other words, a routine for determining a continuous area by extracting only adjacent points in four directions smaller than the threshold value TH, starting from the measurement point P (I, J) first specified by the user, is shown in FIG. This is a check routine.

According to the object extracting method of the third embodiment, it is not necessary to investigate the continuity of all the measurement areas, and only the continuous area including the point designated by the operator can be efficiently extracted. Thus, high-speed object extraction processing can be performed without performing extra continuity comparison processing.

In the third embodiment described above, the four-linking method is used in which the four adjacent points of up, down, left, and right are compared.
Performs not only the upper, lower, left and right of the measurement point but also the comparison processing with the upper left, lower left, upper right, and lower right adjacent measurement points 8
It is also possible to perform continuous surface extraction using the connection method.

Further, the four-link method described in the third embodiment can be used for labeling the entire measurement area in the first and second embodiments. For example, one continuous region: Obj1 is extracted by executing the flow shown in FIG. 12 from one starting point P (I1, J1) in a certain object in the distance image.

Further, a point not included in the extracted continuous area: Obj1 is defined as a starting point P (I
2, J2), and from the starting point P (I2, J2),
By newly executing the flow shown in FIG. 12, another new continuous area: Obj different from the continuous area: Obj1
2 is extracted. By repeating these processes, continuous area extraction of all measurement ranges included in the distance image is executed. Further, if these continuous region extraction processes are performed in parallel, that is, by selecting and executing a plurality of starting points, it is possible to perform continuous surface extraction at higher speed.

The present invention has been described in detail with reference to the specific embodiments. However, it is obvious that those skilled in the art can modify or substitute the embodiment without departing from the spirit of the present invention. That is, the present invention has been disclosed by way of example, and should not be construed as limiting. In addition, aspects in which the embodiments are combined with each other are also included in the gist of the present invention. In order to determine the gist of the present invention, the claims described at the beginning should be considered.

[0121]

As described above in detail, according to the present invention,
For example, based on the three-dimensional information obtained from the three-dimensional sensor, reduce the manual work of the worker in the work of extracting a specific object existing in the three-dimensional space, automatically recognize the continuous surface and efficiently An object extraction device capable of extracting an object,
An object extraction method and a program providing medium can be provided.

Further, according to the present invention, the continuity of the adjacent points is sequentially determined by comparing the distance between the measurement points existing at the adjacent measurement points in the three-dimensional space with a threshold, and the area division is automatically performed. Since it is configured to execute, it is possible to provide an object extraction device, an object extraction method, and a program providing medium, which can extract an object without bothering a worker.

Further, according to the present invention, starting from an arbitrary point on an image, the distance between measurement points existing at adjacent measurement points is compared with a threshold value, the continuity of adjacent points is determined, and only the continuous surface is determined. Since it is configured to extract and sequentially determine only the continuous area and to extract only the continuous surface, it is possible to omit the continuity determination of the discontinuous unnecessary area, and it is possible to perform high-speed object Object extraction device capable of extraction,
An object extraction method and a program providing medium can be provided.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining a conventional object extraction method using a distance image.

FIG. 2 is a diagram illustrating an example of arrangement of a plurality of objects and a three-dimensional sensor arranged in a three-dimensional space, and correspondence between distance sections.

FIG. 3 is a diagram illustrating an extraction mode of an object when various distance sections are set in the object arrangement configuration of FIG. 2;

FIG. 4 is a block diagram mainly showing a functional configuration of a first embodiment in the object extracting apparatus of the present invention.

FIG. 5 is a diagram illustrating a stereo method applicable as a three-dimensional information acquisition configuration used in the object extraction device of the present invention.

FIG. 6 is a diagram illustrating a method of extracting a continuous surface according to the first embodiment in the object extraction device of the present invention.

FIG. 7 is a diagram illustrating a processing flow of continuous surface extraction of the first embodiment in the object extraction device of the present invention.

FIG. 8 is a diagram showing an example of an image obtained as a result of labeling processing by continuous surface extraction in the object extraction device of the present invention.

FIG. 9 is a block diagram illustrating a hardware configuration example of a first embodiment in the object extraction device of the present invention.

FIG. 10 is a diagram illustrating a method of extracting a continuous plane according to a second embodiment in the object extraction device of the present invention.

FIG. 11 is a diagram (part 1) illustrating a processing flow of continuous surface extraction of the second embodiment in the object extraction device of the present invention.

FIG. 12 is a diagram (part 2) illustrating the processing flow of continuous surface extraction in the second embodiment of the object extraction device of the present invention.

[Explanation of symbols]

 401 ... Continuous plane extraction unit 402 ... Target area selection unit 901 ... Computer 902 ... CPU 903 ... ROM 904 ... RAM 905 ... Display interface 906 ... Input device interface 907 ... Hard disk drive 908 ... Display unit 909 ... Mouse 910 ... Keyboard

Continued on the front page (72) Wei Kure, Inventor 6-7-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Inside Sony Corporation (72) Inventor Takayuki Ashigahara 6-35, Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation F-term (reference) 2F065 AA04 AA21 AA53 BB05 FF05 FF09 JJ03 JJ05 JJ26 PP01 QQ00 QQ23 QQ24 QQ25 QQ28 SS02 SS13 2F112 AC06 BA05 CA08 FA19 FA50 5B057 DA06 DB03 DB05 DB09 DC02 DC22 5L09 FA

Claims (21)

[Claims] An object extracting apparatus for extracting an object constituting a continuous surface area from an image based on three-dimensional information, comprising: The point-to-point distance information is calculated, a comparison process is performed between the calculated point-to-point distance information and a predetermined threshold value, and based on the determination that the point-to-point distance information is smaller than the threshold value, An object extraction apparatus, comprising: a continuous surface extracting unit that identifies adjacent measurement points as a continuous surface region having continuity. 2. The apparatus according to claim 1, wherein the distance information between two points is a three-dimensional distance calculated from three-dimensional coordinates of the two adjacent measurement points obtained based on the three-dimensional information. 2. The object extraction device according to 1. 3. The distance information between two points is a depth direction distance which is a distance in a Z-axis direction calculated from three-dimensional coordinates of the two adjacent measurement points obtained based on the three-dimensional information. The object extraction device according to claim 1, wherein: 4. The continuous plane extracting means has a configuration for executing a labeling process for giving a label for identifying a different continuous plane area in correspondence with the measurement point. 2. The object extracting apparatus according to claim 1, wherein a labeling process for giving the same label to each of the identified two measurement points is performed. 5. An image according to claim 1, wherein said image is a distance image obtained based on three-dimensional information. The continuity between the measurement point and the upper adjacent measurement point is determined based on a comparison process between the two-point distance information DU with the measurement point and the threshold, and the left adjacent measurement adjacent to the measurement point in the left direction is performed. The continuity between the measurement point and the left adjacent measurement point is determined based on a comparison process between the point-to-point distance information DL and the threshold value, and only one of the point-to-point distance information DL and DU is determined. Is smaller than the threshold value, the same label as an adjacent measurement point smaller than the threshold value is given as a label of the measurement point, and any of the distance information between two points DL and DU is used. Is smaller than the threshold value, a new label is given as the label of the measurement point. If both of the distance information DL and DU between the two points are smaller than the threshold value, A measurement point to which one of the identification labels is selected as a selection label and is provided as a label of the measurement point, and the same label as the unselected label in any of the upward and left directions different from the selection label is provided. 2. The object extracting apparatus according to claim 1, wherein the apparatus has a configuration for executing a process of changing a label to a selected label. 6. An image according to claim 6, wherein said image is a distance image obtained based on three-dimensional information. Point-to-point distance information DU with a point, Point-to-point distance information DL with an upper adjacent measurement point adjacent to the measurement point in the left direction, Point-to-point measurement with an upper adjacent measurement point adjacent in the downward direction of the measurement point The distance information DD, the distance information DR between two points with the upper adjacent measurement point adjacent to the measurement point to the right in the right direction, and the adjacent measurement in the four directions of up, down, left, and right based on the comparison processing with each of the threshold values The continuity with each point is determined, and the adjacent measurement point determined to have continuity is set as the next measurement point. Object extracting apparatus according to claim 1, characterized in that the fourth connecting method of repeating the process for sequentially with a configuration to perform continuous surface extraction. 7. The continuous plane extracting means has a configuration in which an arbitrary point of an image representing an object to be extracted is set as a starting point of the measurement point, and continuity with an adjacent measurement point is sequentially determined to perform continuous plane extraction. The object extraction device according to claim 6, wherein: 8. The continuous plane extracting means determines continuity of a measurement point included in a distance image obtained based on three-dimensional information with eight adjacent measurement points around the measurement point, and determines that there is continuity. Claims are characterized in that a continuous face extraction is performed by an 8-connection method in which a determined adjacent measurement point is set as a next measurement point and a process of determining continuity with eight neighboring measurement points is sequentially repeated. Item 1. The object extraction device according to Item 1. 9. The continuous plane extracting means has a configuration in which an arbitrary point of an image representing an object to be extracted is set as a starting point of the measurement point, and continuity with adjacent measurement points is sequentially determined to perform continuous plane extraction. The object extraction device according to claim 8, wherein: 10. The object extraction method according to claim 1, wherein each of the measurement points is associated with a pixel included in the image, and adjacent measurement points correspond to adjacent pixels. apparatus. 11. An object extraction method for extracting an object constituting a continuous surface area from an image based on three-dimensional information, the method comprising the steps of: A point-to-point distance information calculating step of calculating point-to-point distance information; a comparison processing step of executing a comparison process between the calculated point-to-point distance information and a predetermined threshold value; A step of identifying the adjacent measurement points as a continuous plane area having continuity based on the determination that the measurement points are smaller than each other. 12. The two-point distance information used in the two-point distance information calculating step is a three-dimensional distance calculated from three-dimensional coordinates of the two adjacent measurement points obtained based on the three-dimensional information. The object extraction method according to claim 11, wherein 13. The two-point distance information used in the two-point distance information calculating step is a Z-axis direction calculated from three-dimensional coordinates of the two adjacent measuring points obtained based on the three-dimensional information. The method according to claim 11, wherein the distance is a depth direction distance. 14. The object extracting method according to claim 1, further comprising a labeling processing step of assigning a label for identifying a different continuous plane area to the measurement point, wherein the labeling processing step is a step of identifying the continuous plane area identification step. The object extraction method according to claim 11, wherein the same label is assigned to each of the two measurement points identified as a continuous surface area having continuity. 15. The point-to-point distance information calculating step includes the step of calculating point-to-point distance information between a measurement point and an upper adjacent measurement point that is adjacent in the upward direction with respect to the measurement point included in the distance image obtained based on the three-dimensional information. Calculating the DU; and calculating the distance information DL between two points between the measurement point and the left adjacent measurement point adjacent in the left direction. The comparison processing step includes: Executing a comparison process between the two-point distance information DL and the threshold value, and performing a comparison process between the two-point distance information DL and the threshold value. And determining the continuity between the measurement point and the left adjacent measurement point, and the upper adjacent measurement point based on The continuity between the measurement point and the upper adjacent measurement point is determined based on a comparison process between the two-point distance information DU between the measurement point and the upper adjacent measurement point adjacent to the direction and the threshold value. The continuity between the measurement point and the left adjacent measurement point is determined based on a comparison process between the distance information DL between two adjacent left measurement points and the threshold value, and the distance information between two points DL, If only one of the DUs is smaller than the threshold, the same label as the adjacent measurement point smaller than the threshold is given as the label of the measurement point, and both of the distance information between two points DL and DU are If it is not smaller than the threshold value, a new label is given as the label of the measurement point. If both of the distance information DL and DU between the two points are smaller than the threshold value, One of the identification labels in the upward direction and the left direction is used as the selection label, and the label is attached to the measurement point, and the same label as the non-selection label in the upward or left direction different from the selection label is used. 12. The method according to claim 11, wherein a process of changing the label of the measurement point to which the symbol is added to the selected label is executed. 16. The point-to-point distance information calculating step includes the step of calculating a distance between two points of a measurement point included in a distance image obtained based on three-dimensional information with an upper adjacent measurement point adjacent in the upward direction of the measurement point. Information DU, distance information DL between two points with an upper adjacent measurement point adjacent to the measurement point in the left direction, distance information DD between two points with an upper adjacent measurement point adjacent in a downward direction of the measurement point, the measurement point The distance information DR between two points with the upper adjacent measurement point adjacent in the right direction is calculated, and the comparison processing step calculates the distance information 2 calculated in the distance information calculation step between two points.
A comparison process is performed between each of the point-to-point distance information DU, DL, DD, and DR and the threshold value, and the continuous plane region identification step includes: Determines the continuity with the measurement point, sets the adjacent measurement point determined to have continuity as the next measurement point, and further repeats the process of sequentially determining the continuity with each of the adjacent measurement points in four directions, up, down, left, and right. 2. The method according to claim 1, wherein a continuous surface is extracted by a method.
12. The object extraction method according to item 11. 17. A starting point setting step of setting an arbitrary point of an image indicating an extraction target object as a starting point of the measurement point, wherein the step of calculating distance information between two points includes the step of setting the point set in the starting point setting step. 17. The object extraction method according to claim 16, wherein continuity extraction is performed by successively determining continuity with adjacent measurement points by setting as a first measurement point. 18. The object extraction method according to claim 1, wherein the continuity of the measurement point included in the distance image obtained based on the three-dimensional information is measured with eight adjacent measurement points around the measurement point, and the continuity is determined. 12. The continuous surface extraction is performed by an 8-connection method in which the determined adjacent measurement point is set as the next measurement point and the process of judging the continuity with eight adjacent measurement points is sequentially repeated. Object extraction method. 19. A starting point setting step of setting an arbitrary point of an image representing an extraction target object as a starting point of the measurement point, wherein the step of calculating distance information between two points includes the step of setting the point set in the starting point setting step. 19. The object extraction method according to claim 18, wherein the first measurement point is used as a first measurement point, and continuity with an adjacent measurement point is sequentially determined to perform continuous plane extraction. 20. The object extraction method according to claim 11, wherein each of said measurement points is associated with a pixel included in said image, and adjacent measurement points correspond to adjacent pixels. 21. A program providing medium for tangibly providing a computer program for causing a computer system to execute a process for executing an object extraction process based on three-dimensional information, wherein the computer program comprises: Calculating a distance information between two points between two adjacent measurement points in the measurement points included in the three-dimensional information; and comparing the calculated distance information between the two points with a predetermined threshold value A comparison processing step of executing processing; and a continuous surface area that identifies the adjacent measurement points as a continuous surface area having continuity based on the determination that the distance information between two points is smaller than the threshold. A program providing medium, comprising: an identification step.